Errors

Status codes

Code	Meaning	Retry?
422	The request body is wrong. Offending fields are listed	Only after fixing the body
409	Well-formed, but a precondition failed	No — the same request produces the same conflict
404	Does not exist for your organisation. Also the answer when it exists but is someone else’s	No
403	Your role does not permit this	No
503	An upstream is temporarily unavailable	Yes, with backoff

The 409 case is the one to internalise. It is not “try again later” — it is “the world is not in a state where this can succeed”, and the message says which state.

Messages you are likely to meet

“…subnet’inin internet çıkışı yok” — the subnet has no gateway. Appliances fetch their own configuration and cannot be created without egress. Attach a gateway.

“data katmanı: izole kalır, geçide takılamaz” — you tried to attach a gateway to a data subnet. That tier can never have egress. Use app or mgmt.

“Organizasyonunuza tahsis edilmiş boş public IP yok” — no address is allocated to you and free. An address attached to a machine is not free. Ask your operator for one.

“Boştaki N adresin hiçbiri kullanılamıyor” — you have addresses but none can be used: they are the pool’s own gateway, or network/broadcast addresses of a subnet-kind pool. Different from the message above, deliberately — you can see them in the list, so being told “you have none” would be wrong.

“bir portun bütün dinleyicileri aynı protokolde olmalı” — you are adding an https listener to a port that already has a tcp one, or the reverse.

“…portunda zaten bir dinleyici var” — that port and hostname combination is taken, or you left the hostname empty on a port that already has a default.

“alan adının DNS’i bizde değil” — the domain is in your account but its DNS is hosted elsewhere, so the certificate challenge record cannot be written. See Register a domain and host its DNS.

“Bu dinleyici … deposunun uç noktası” — the listener belongs to an object storage store. Change it from the storage screen; editing it here would leave the store with an address nothing serves.

“Yük dengeleyici başka bir özel ağda” — a load balancer has one leg and reaches only its own network. Create one in the right network.

“Altyapı kuruldu ama N düğüm Kubernetes’te hazır değil” — the machines exist but did not become ready. Almost always the subnet has no gateway, so images cannot be pulled. Fix it, then re-verify.

“Küme API’sine ulaşılamadı” — the panel could not reach the cluster to measure it. Check the cluster is running and the management path is intact.

Job failures

A job that reaches failed records its error. The resource carries the reason too.

Errors are classified: a temporary failure — a busy hypervisor, a held lock, an upstream timeout — puts the job into retry and the resource is not marked failed, because showing “error” for a transient condition pushes people to delete and rebuild something that was about to succeed. A permanent failure stops the job.

When you think the panel is wrong

Before filing a bug, check Intent, truth and reconciliation. The most common “the panel is wrong” is a status that is only measured when something runs. Where that is the case there is an explicit way to ask for a fresh measurement — the Kubernetes re-verify action is the clearest example — and it is separate from repairing anything.